feat(gkd): add agentic monitoring metrics - #10026
Open
Casuallkk wants to merge 1 commit into
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
PR type
PR information
Add monitoring metrics for agentic multi-turn GKD training:
num_turns: average number of turns in on-policy trajectories.gkd/topk_overlap: mean token-level overlap between teacher and student top-k predictions.gkd/teacher_student_gap: mean teacher-student log-probability difference on observed response tokens, defined asE_t[log p_teacher(y_t) - log p_student(y_t)].The metrics are supported across Hugging Face, Megatron, and Ray-Megatron GKD paths. For teacher API top-k outputs, the observed token log-probability is preserved even when that token falls outside the retained teacher top-k.
Experiment results
pytest tests/train/test_gkd_monitoring.py -q: 4 passed.